Study of early termination for clinical trials will continue using analytical and simulation approaches. Work will concentrate on the distribution of the number of events until termination is first indicated assuming early rejection will be stochastic, i.e. will occur if with high probability, under Ho, Ho will be rejected while early acceptance will be deterministic. This work will assume death as an end point, concurrent entry of all participants in the trial and that the statistic of interest is the Wilcoxon. Several exprimental designs will be considered including follow-up until all deaths have occurred, until a specified percent of deaths has occurred, in either group. It is anticipated that the first two settings are more likely to permit useful results. Problems suggested by contacts during the last year with the Lipids Research Clinics Central Patient Registry (University of North Carolina) will be pursued to completion. These include a paper in approaches to data-monitoring in a clinical trials and a paper comparing the sensitivity of two tests of association under multivariate normal assumptions. One is the usual test of sample R2. The other test is a Hotelling T2 test of the sample x vectors for the 100 theta percent largest y's in a sample with the sample x vectors for the 100 theta percent smallest y's. In addition, revision of papers in various stages of the publication process will be pursued.